Bull Volcanol (2016) 78: 72 DOI 10.1007/s00445-016-1062-z

RESEARCH ARTICLE

Constraining pre-eruptive magma conditions and unrest timescales during the Monte Nuovo eruption (1538 AD; Campi Flegrei, Southern ): integrating textural and CSD results from experimental and natural trachy-phonolites

Fabio Arzilli1,2 & Monica Piochi3 & Angela Mormone3 & Claudia Agostini2 & Michael R. Carroll2

Received: 10 December 2015 /Accepted: 3 September 2016 /Published online: 24 September 2016 # The Author(s) 2016. This article is published with open access at Springerlink.com

Abstract We present crystallization experiments representing The crystallization time of the magma requires that it ascended a broad range of growth conditions of alkali feldspar and so- from pre-eruptive storage to the quenching level in several dalite in a trachy-phonolite magma composition during later hours to a few days. We also observe that a small decrease stages of evolution. Our results include (i) textural data and in pressure can induce a dramatic increase in crystallinity, with mineral assemblages of synthetic samples; (ii) feldspar nucle- associated rheological changes, leading to changes in the ation kinetics and growth rate estimates; and (iii) textural data, eruption style, and such changes could occur on timescales mineral abundances, and crystal size distribution measure- of hours to several days. The products from the later phases ments on samples representative of the Monte Nuovo eruption of the Monte Nuovo eruption are more crystalline and contain (1538 AD), the last eruption of Campi Flegrei, Southern Italy. sodalite in response to the decrease in magma ascent rate, Experiments reproduced the texture and feldspar content of whichinturnallowedformoredegassingduringascent, natural products indicating that kinetic data can provide in- resulting in more time spent at very shallow depths. sights into processes within the volcanic system shortly before and during this small-magnitude eruption and, particularly, Keywords Alkali feldspars . Trachytic melts . Crystallization about magma ascent timescale. We suggest that the ground- kinetics . CSD . Monte Nuovo . Campi Flegrei mass crystallization of Monte Nuovo magma started between 4 and 7 km depth (∼100–200 MPa) at a temperature between 825 and 840 °C (close to the liquidus of alkali feldspar). The Introduction crystallization kinetics of alkali feldspar and the absence of sodalite in most of the natural samples indicate that magma The aim of the study ascent rate increased in the shallow part of the conduit from about 3 km depth to the quenching level (possibly fragmenta- Alkali feldspar is an abundant phase in evolved alkaline rocks ∼ tion point; 30 MPa), during the first phases of the eruption. (phonolites, trachytes) and is widespread in Province magmas (e.g., Piochi et al. 2005). Its occurrence as Editorial responsibility: P. Wallace microlites in pumice and scoria provides potential information on the timescale of magma migration within the crust, specif- * Fabio Arzilli ically from the magma chamber to the fragmentation level [email protected] (Marsh 1988; Cashman and Marsh 1988;Marsh1998). This information may be unraveled by studying the size distribu- 1 School of Earth and Environmental Sciences, The University of tion of microlites pre- and syn-eruptively crystallized and Manchester, Oxford Road, Manchester M13 9PL, UK quenched in the matrix of juvenile magma fragments. The 2 School of Science and Technology–Geology Division, University of crystals present in an igneous rock and the observed variations Camerino, Via Gentile III da Varano, 62032 Camerino, Italy in both their composition and texture reflect the integrated 3 Istituto Nazionale di Geofisica e Vulcanologia, sezione Osservatorio pressure (P)–temperature (T)–composition (X)–time (t)histo- Vesuviano, Via Diocleziano 328, 80124 , Italy ry of the magma from which they formed. As a result, it is 72 Page 2 of 20 Bull Volcanol (2016) 78: 72 possible to link textural observations on rocks with experi- understanding how fast the crystallinity of a trachytic mentally derived data for rates of crystal growth (YL) and magma could change after small variation of pressure. crystal number density (Na) for specific mineral phases and These experiments provide constraints on the P–T condi- the undercooling (ΔT) values of the parental melt. This ap- tions of trachytic to phonolitic melts during their ascent proach makes it possible to obtain information on magmatic to the surface, and we use the results to make interpre- processes and their timescales using textural observation and tations about volcanological and magmatic processes dur- growth rates of crystals (Cashman and Marsh 1988;Brugger ing the Monte Nuovo eruption (1538 AD)intheCampi and Hammer 2010a; Eberl et al. 2002). Flegrei (Southern Italy; Fig. 1). However, there are few studies about crystallization kinetics The Monte Nuovo eruption is particularly interesting be- on trachy-phonolitic melts (e.g., Iezzi et al. 2008;Arzilliand cause it occurred after 3000 years of volcanic quiescence and Carroll 2013). In this study, we present results for 16 new crys- following a period of ground level movements and seismicity tallization experiments on hydrous trachytic melt. These data (as described in historical chronicles; Guidoboni and complement the previous experimental work of Arzilli and Ciuccarelli 2011) that have characteristics comparable to the Carroll (2013) on the same starting composition. We performed recent bradyseisms in the Campi Flegrei area (Parascandola isobaric cooling, isothermal decompression and decompression 1947; Del Gaudio et al. 2010). Constraining the timing of + cooling experiments, investigating experimental durations magma movements in the subsurface shortly before the erup- and P–T conditions not already investigated by Arzilli and tion is a key factor to unraveling the significance of phenom- Carroll (2013). All of the experiments use the single-step meth- ena affecting the Campi Flegrei during volcanic quiescence. od to reproduce the trachytic meltevolutioninresponsetoan Previous textural studies on the Monte Nuovo eruption instantaneously applied thermodynamic driving force (i.e., (D’Oriano et al. 2005;Piochietal.2005)werehamperedby undercooling, ΔT = Tliquidus − Texperimental). In our work, we the lack of crystallization kinetic data for trachytic-phonolitic have not attempted to distinguish between the effects of melt compositions. To constrain the conditions and the time- undercooling (ΔT) at constant pressure and effective scales of magmatic processes, we focus on the crystallization undercooling (ΔTeff), which is caused by decompression at con- kinetics of alkali feldspar in trachytic melts and new crystal stant temperature (Hammer and Rutherford 2002), because it size distribution (CSD) data for groundmass feldspars in the has been shown by Shea and Hammer (2013) and Arzilli and natural samples. This choice, although empirical (Brugger and Carroll (2013) that under the same final temperature and pres- Hammer 2010a; Eberl et al. 2002), is in line with several sure, both processes appear to produce similar nucleation and studies (e.g., Geschwind and Rutherford 1995;Hammerand growth rates. Furthermore, we performed decompression + Rutherford 2002; Couch et al. 2003; Armienti et al. 2007; cooling experiments, in which it is difficult to separate ΔT and Martel 2012) that used crystallization kinetics and CSD of

ΔTeff. Adiabatic cooling during magma ascent could be impor- feldspar (mainly plagioclase) to unravel shallow dynamics at tant, but the magnitude of this effect could be partially several calc-alkaline volcanoes. Our new data (including counteracted by the release of latent heat from crystallization potential complications regarding interpretations of natural (La Spina et al. 2015). Decompression + cooling experiments textures, e.g., Brugger and Hammer 2010a;Eberletal. were performed with the goal of studying the crystallization of 2002) provide a quantitative method (in the view of Brugger magma resulting from degassing and adiabatic cooling during and Hammer 2010a) for studying magma dynamics using the ascent. The decompression + cooling experiments also allowed abundant alkali feldspar crystals in the Campi Flegrei system. us to investigate several final pressures and undercoolings that could not be studied by simple isothermal decompression. The experiments were conducted at different pressures Volcanological background and ΔT values to highlight the differences in crystallinity between high and low pressure (and thus melt H2Ocon- The Monte Nuovo eruption tent). The preliminary study of Arzilli and Carroll (2013) focused mainly on the influence of ΔT and time on the The 1538 Monte Nuovo eruption (Fig. 1a–e) was character- nucleation and growth of alkali feldspar. They showed ized by relatively low intensity and magnitude (Di Vito et al. the occurrence of several nucleation events of alkali feld- 1987;D’Oriano et al. 2005; Piochi et al. 2005, 2008). Similar spar in short times (hours), with dominance of nucleation to the majority of volcanic eruptions of the last 14.9 ka BP in at large ΔT. Here we utilize our new experiments and thearea(seeRosietal.1983;DiVitoetal.1999; Isaia et al. those performed by Arzilli and Carroll (2013)toinvesti- 2009), it occurred from a monogenic vent and released gate how crystallization may change as a function of <0.1 km3 of magma (Lirer et al. 1987;RosiandSbrana

PH2O (proportional to melt water content for the water- 1987; Mormone et al. 2011) with a K-rich, phono-trachytic saturated conditions investigated) and ΔT (induced by composition (Fig. 2). The erupted material consisted of vari- cooling and/or decompression). Our study is focused on ably vesicular pumice and scoria fragments with sparsely Bull Volcanol (2016) 78: 72 Page 3 of 20 72

Fig. 1 The Monte Nuovo tuff cone location (a) with image of the lowermost ash to sand levels and the uppermost scoria deposits (b)and(c), respectively. The diagrams show the timing of the phenomena preceding the eruption since 1470 AD (d)and 1538 (e), on the basis of Guidoboni and Ciuccarelli (2011). The panels (d)and(e) indicate the time of occurrence for seismicity (vertical bars), flow output (cloud), and ground deformation (arrows and dotted lines; right concavity inflation, left concavity deflation). The relation between phenomena and outcrops is also indicated. The onset for microlite growth as suggested by CSD is also indicated. The brown pumice and scoria samples’ location is indicated by black boxes in panels (a)and(b). The sampling site of the early erupted yellow pumices of LM inf d and LM inf l is indicated in panel (a)

porphyritic texture and typically containing alkaline feldspars the past 14.9 ka, Campi Flegrei volcanism mostly concentrat- (e.g., Rosi and Sbrana 1987;Piochietal. 2008;D’Oriano et al. ed in epochs of intense activity, alternating with quiescent 2005). However, the Monte Nuovo eruption seems to be an periods of thousands of years; in such epochs eruptions oc- exceptional event in the Campi Flegrei history, given that in curred with a decadal frequency (Di Vito et al. 1999; Di Renzo

Fig. 2 TAS (Le Bas et al. 1986), on the left, and Alkali vs. CaO, on the right, diagrams for the Monte Nuovo natural glasses and ZAC bulk rock (Table 1). Data on natural glass and whole rocks from Piochi et al. (2008), Rosi and Sbrana (1987), D’Oriano et al. (2005) 72 Page 4 of 20 Bull Volcanol (2016) 78: 72 et al. 2011). Instead, the Monte Nuovo event was the last and same alignment. Overall, the products are denser and more occurred after a volcanic quiescence of ∼3000 years. crystalline upwards in the sequence; at the same time, the The eruption started in the night between September 29th temperatures suggested by ternary solvus feldspar microlites and 30th (Guidoboni and Ciuccarelli 2011), after an intensifi- moves to higher values (900 to 1000 °C), as observed by cation of decadal fumarolic and bradyseism activity D’ Oriano et al. (2005) and Piochi et al. (2005). Piochi et al. (Parascandola 1947; Lirer et al. 1987; Guidoboni and (2008) also measured low water content in the glassy ground- Ciuccarelli 2011) occurred about 3 months before the eruption mass of pumice and scoriae; therefore, degassing was consid- (Guidoboni and Ciuccarelli 2011; Fig. 1d, e). The first ered the driving force for crystallization. Measured CSDs dis- phreato-magmatic phase produced pumiceous lapilli-bearing play concave upward trends attributed to two crystal popula- ash deposits (LM in D’Oriano et al. 2005;UnitsI–II in Di Vito tions (D’Oriano et al. 2005; Piochi et al. 2005; Mastrolorenzo et al. 1987) that built the cone. The main LM deposit is com- and Pappalardo 2006). posed of coarse to fine ash and is coarser and lapilli bearing in the lowermost levels (Unit I), where it is intercalated with pumice or scoria lenses or beds (Boivin et al. 1982). This deposit was emplaced by pyroclastic flows. On October 3, Methods: synthetic and natural products after 2 days of diminishing intensity, explosions produced two dark scoriae layers intercalated with a gray ash bed Synthetic products: starting material and experimental (UM1 or Unit III, in D’Oriano et al. 2005 and Di Vito et al. strategy 1987, respectively). The final phase, after 3 days of eruptive quiescence, produced a scoria unit deposited by pyroclastic We complement the experiments of Arzilli and Carroll (2013) flow (UM2 in D’Oriano et al. 2005) and/or fallout (Unit IV to obtain a wider data set of feldspar crystallization conditions in Di Vito et al. 1987)mechanisms.D’Oriano et al. (2005) involving growth under isobaric cooling, isothermal decom- suggested that the UM1 scoria-bearing pyroclastic flows were pression, and decompression + cooling experiments. The originated by vulcanian-type explosions. starting material (Table 1) is the same obsidian used in previ- ous work; it was sampled within the Breccia Museo Unit, a Overview of previous studies pyroclastic breccia associated with the Campanian Ignimbrite eruption dated at c.a. 39 ka (Melluso et al. 1995; Fedele et al. The pumice and scoria fragments contain <3 % phenocrysts, 2008). The sample is known as ZAC in the literature (Di consisting mostly of feldspars, with subordinate magnetite and Matteo et al. 2004; Fabbrizio and Carroll 2008; Calzolaio rare spinel, and are microcrystalline, with a variety of textures et al. 2010; Arzilli and Carroll 2013). It contains <10 % crys- (D’Oriano et al. 2005;Piochietal.2005, 2008). Pumices are tals by volume. Phenocrysts are mostly alkali feldspar with mostly angular clasts, yellowish in color and with sub- minor plagioclase, clinopyroxene, biotite, and magnetite, in spherical vesicles, from 2 to 30 cm in dimension (Di Vito order of decreasing abundance; apatite and titanite are acces- et al. 1987). Other pumices are characterized by alternating sory phases. The composition is trachytic, near the trachyte- yellow and brown portions and are referred to as banded pum- ices. Scoriae are mostly blackish clasts, generally decimeter- – Table 1 Chemical composition of ZAC (from Di Matteo et al. 2004) sized (20 60 cm diameter; Di Vito et al. 1987), containing and average for natural samples from units selected for this study (see heterogeneous vesicles and often a more vesicular core and Piochi et al. 2008 for original, not H2O-free normalized data) less vesicular margin, with elongated vesicles in some pumice Oxide (wt%) ZAC LM inf LMc MN4 fragments. Microlites are primarily alkali feldspars (sanidine and lesser anorthoclase), and microlite abundance varies with SiO2 62.18 60.76 60.77 63.71 vesicularity. Microlites in highly vesicular pumice are less TiO2 0.45 0.42 0.43 0.31 abundant and smaller than those in scoria clasts. Banded pum- Al2O3 18.70 19.09 19.39 19.91 ice fragments are characterized by strong parallel layering and FeO* 3.19 3.05 3.05 1.63 variations in groundmass texture: brown bands are more crys- MnO 0.27 0.27 0.27 0.07 talline and less vesicular than yellowish bands, which are sim- MgO 0.23 0.23 0.24 0.12 ilar to the yellowish pumice (Piochi et al. 2008). The CaO 1.65 1.89 1.94 1.55 microlites of banded pumices have generally acicular shape Na2O 6.16 7.09 6.82 6.67 with larger sizes compared with those in scoriae; some of them K2O 7.14 7.17 7.08 6.04 are curved, following the shape of an adjacent vesicle, sug- P2O5 0.02 0.03 0.02 0.01 gesting that they grew concomitantly with vesicle formation Total 100.00 100.00 100.00 100.00 (Piochi et al. 2005, 2008). In pumices, microlites are weakly to strongly aligned, and microlites and color bands share the FeO* total iron as FeO Bull Volcanol (2016) 78: 72 Page 5 of 20 72 phonolite border in the TAS diagram (Di Matteo et al. 2004). and experimental durations (see Table 2)relativetothoseper-

It is similar but not identical to the Monte Nuovo products in formed by Arzilli and Carroll (2013). The amount of H2O terms of whole-rock and glassy matrix chemistry (Fig. 2; needed to achieve water saturation was estimated using the

Table 1). There are also small differences in phenocryst con- H2O solubility data of Di Matteo et al. (2004). The initial tents (<10 % ZAC vs. 3 % Monte Nuovo) and mineral assem- conditions of the experiments were above the liquidus of the blage (slight abundance of pyroxene and biotite in ZAC). alkali feldspar. The experiments were carried out using cold Despite these small differences, we used ZAC because the seal pressure vessels (CSPV) at the Geology Division phase diagram (Fabbrizio and Carroll 2008; Arzilli and (University of Camerino), using Nimonic 105 alloy cold-seal Carroll 2013) allowed us to choose accurately the experimen- bombs. The bombs were pressurized using water as the pres- tal conditions relevant for our study on the feldspar crystalli- sure medium. The redox conditions of the apparatus are ∼0.8 zation kinetic. In addition, Piochi et al. (2008) proposed a log fO2 units above the NNO buffer (Di Matteo et al. 2004). similar composition for the magma systems feeding many The temperature was measured with a K-type (chromel- eruptions at the Campi Flegrei. alumel) thermocouple, with an accuracy of ±5 °C. Cold seal To complement the experimental data of Arzilli and Carroll pressure vessels were equipped with rapid quench extensions, (2013), we have performed 16 new crystallization experi- allowing samples to be dropped into a water-cooled chamber ments. Particularly, we extend cooling and degassing experi- at the end of the experiments (Arzilli and Carroll 2013). The ments in water-saturated conditions towards higher pressure samples were cooled, reaching ambient temperature in a few

(experimental final pressure, Pf = 100, 150, and 200 MPa) and seconds, with quench rates of ∼150 °C/s (Dobson et al. 1989; lower pressure (Pf = 30 and 50 MPa) within a range of tem- Carroll and Blank 1997). Capsules (Ag70-Pd30) were filled peratures between 750 and 850 °C (Table 2). We performed with ZAC powder and H2O (slightly more than the amount isobaric cooling and isothermal decompression experiments of H2O needed to reach the saturation condition) and then with experimental durations not already investigated by sealed to obtain a closed system. Capsules were weighed after Arzilli and Carroll (2013). In addition, decompression + each experiment to check for changes in weight, and capsules cooling experiments present different initial temperatures with no weight change were deemed successful.

Table 2 Experimental conditions for isothermal cooling, decompression + cooling and isothermal decompression experiments

* Sample Ti (°C) Tf (°C) P (MPa) tm (s) texp (s) H2O(wt%) ΔT (°C) –ΔT (°C)

Isobaric cooling experiments D20 900 750 200 72,000 28,800 7 20 131 D15 900 775 150 7200 7200 6 17 108 D13 880 775 150 7200 28,800 6 17 88 D6 880 800 100 7200 21,600 5.2 19 61 D11 900 850 50 7200 7200 2.7 15 35 D12 900 850 50 7200 14,400 2.7 15 35 D33 900 840 50 7200 14,400 2.7 25 35 D10 900 825 50 7200 28,800 2.7 40 35 D9 900 825 50 7200 57,600 2.7 40 35 D69 880 750 50 7200 21,600 2.7 115 15 D78 880 750 50 7200 50,400 2.7 115 15 * Sample Ti (°C) Tf (°C) Pi (MPa) Pf (MPa) tm (s) texp (s) H2O(wt%) ΔT (°C) –ΔT (°C) Decompression + cooling experiments D17 825 775 200 150 72,000 21,600 6 17 56 D25 825 775 200 150 72,000 57,600 6 17 56 D29 825 800 200 100 7200 57,600 5.2 19 56 D21 825 825 200 50 7200 57,600 2.7 40 56 Isothermal decompression experiments D62 825 825 150 30 3600 50,400 1.7 65 33

*H2O (wt%) was calculated using the polynomial fit by Di Matteo et al. (2004)

Ti initial (or melting) temperature, Tf final (or experimental) temperature, Pi initial (or melting) pressure, Pf final (or experimental) pressure, tm melting time, texp experimental time, ΔT the undercooling degree, −ΔT the superheating degree 72 Page 6 of 20 Bull Volcanol (2016) 78: 72

Natural products products higher in the sequence. In contrast, the MN4top sco- riae are interpreted to have formed by degassing of magmatic Analyzed pumice and scoria fragments are from the Lower and volatiles during rapid ascent and resultant magma fragmenta- Upper Members of the Monte Nuovo formation (Fig. 1a–c), tion (D’Oriano et al. 2005;Piochietal.2008). The MN4top respectively. Fragments and eruption levels were selected on scoriae have low H2O contents (0.21 ± 0.05 wt%; Piochi et al. the basis of previous textural and chemical characterization 2008). We also consider brown and yellow sectors in banded (D’Oriano et al. 2005;Piochietal.2005;Piochietal.2008) pumices from LM (LMC3; Fig. 1c) in order to examine the with the aim of analyzing in greater detail (including new X-ray entire textural range. diffraction data) the textural and petrological variability in the eruption products. In particular, we focus on pumices from the base of the sequence (samples LM inf d and LM inf l; Fig. 1a) and top scoriae (MN4top; Fig. 1bandFig.3b). The basal pum- Analytical methods ices are interpreted to have formed by fragmentation involving some interaction with external water during magma ascent. SEM and image analyses Textural evidence for this (Fig. 3) includes (i) vesicle walls that are broken and display shock-induced cracks, (ii) higher vesic- Electron microscope images were taken for both synthetic and ularities (up to 80 %) than other pumices and scoriae, and (iii) natural samples in order to obtain detailed information on tex- lower groundmass crystallinity. These pumices have higher tural features and mineralogical assemblage (Figs. 3, 4,and5).

H2O contents (1.11 ± 0.16 wt%; Piochi et al. 2008)than Backscattered electron (BSE) images were collected for

Fig. 3 Groundmass texture of natural pumice (samples c1 1-7, c2s 9-12, infl 8-12, infsc 02-05) and scoriae (samples MN3/1, MN4top) from base upwards in the stratigraphic sequence. Samples MN3/1 (a) and MN4top (b) are two scoriae in the upper part of the sequence. Note the cuspate shape of vesicle and the magnetite occurrence in the most crystalline and poorly vesiculated MN4top. Samples c1 1-7 (c)and c2s 9-12 (d)showthe groundmass crystallinity and vesicularity feature of a yellowish and a brown pumice, respectively, within the main pyroclastic unit. Note the microlite alignment and the curved shape of several crystals. Rare microlites and high vesiculation characterize pumice at the base of the flow sequence. In particular, sample infl 8-12 (e) shows the cracks (white arrows) produced by water/magma interaction. Sample infsc 02-05 (f) is a banded pumice within the main pyroclastic unit characterized by low crystal abundance and higher vesiculation of yellowish band (on the left) with respect of the adjacent brown sector. White bars are 100 μm Bull Volcanol (2016) 78: 72 Page 7 of 20 72

Fig. 4 Backscattered SEM images of textures obtained at low pressure D47 (Pf =30MPa,ΔT = 140 °C). Isothermal decompression (50 and 30 MPa) from isobaric cooling, decompression + cooling and experiments: g D57 (Pf =50MPa,ΔT = 40 °C); h D60 (Pf = 30 MPa, isothermal decompression experiments. Cooling experiments: a D33 ΔT = 65 °C); i D62 (Pf =30MPa,ΔT = 65 °C). The samples D33, D10, (Pf = 50 MPa, ΔT = 25 °C); b D10 (Pf =50MPa,ΔT =40°C);c D69 D69, and D62 were performed in this study; the other ones are data of (Pf =50MPa,ΔT = 115 °C). Decompression + cooling experiments: d Arzilli and Carroll (2013) D45 (Pf = 50 MPa, ΔT =115°C);e D43 (Pf = 50 MPa, ΔT =115°C);f

2 experimental samples using a JEOL JSM-6390LA FE-SEM at cm ). Na and crystal area fraction (ϕ)ofalkalifeldsparwere the Institute of Geochemistry and Petrology, ETH Zurich, and calculated on a vesicle-free basis. The relevant area (Ar)in JEOL JSM-6500F–upgraded to 7000 series–FEG at the INGV each BSE image consists of only glass and crystal phases, of Rome. BSE images of pumices and scoriae of Monte Nuovo thus, all abundances refer to phase proportions in glass plus were acquired using a LEO 430 SEM at BIstituto Geomare crystals (vesicle-free basis; e.g., Hammer et al. 1999). Na of Sud–CNR^ of Naples, Italy and a ZEISS SUPRA 35 at the alkali feldspar was obtained through the relation (Hammer BDipartimento di Ingegneria dell’Informazione^ of the II et al. 1999): Università di Napoli in Aversa, Italy, operating at 10 kV, and a working distance of 15 and 7–12 mm, respectively. nFsp N a ¼ ð1Þ Textural analysis on BSE images was performed using Ar ImageJ software (NIH Image; Abramoff et al. 2004; Schneider et al. 2012) to obtain crystallization kinetics data The alkali feldspar volume fraction (ϕ) was obtained from for alkali feldspar. Nucleation and growth kinetics were cal- (Hammer et al. 1999): culated using the batch method (Hammer and Rutherford AFsp 2002; Couch et al. 2003; Brugger and Hammer 2010a, b). ϕ ¼ ð2Þ A Imaging analyses were performed to measure the number of r alkali feldspar crystals and the area of each phase (crystal where AFsp is the area of the alkali feldspar. Since alkali feld- phases, glass, and bubbles) in every (both synthetic and natu- spar crystals and trachytic glass have similar average atomic ral) sample. The number of alkali feldspar crystals (nFsp)was number, BSE images often showed low contrast between measured to calculate the crystal number density (Na,crystals/ these two phases. Therefore, the separation of alkali feldspar 72 Page 8 of 20 Bull Volcanol (2016) 78: 72

Fig. 5 Backscattered SEM images of textures obtained at high pressure ΔT =17°C);e D55 (Pf =150MPa,ΔT =42°C);f D51 (200, 150, 100, and 70 MPa) from isobaric cooling, decompression + (Pf = 100 MPa, ΔT = 69 °C). Isothermal decompression experiments: g cooling, and isothermal decompression experiments. Cooling D80 (Pf = 100 MPa, ΔT = 19 °C); h D82 (Pf =70MPa,ΔT = 14 °C); i experiments: a D1 (Pf = 200 MPa, ΔT =20°C);b D15 D82 (Pf = 70 MPa, ΔT = 14 °C). The samples D15, D6, and D17 were (Pf = 150 MPa, ΔT =17°C);c D6 (Pf = 100 MPa, ΔT = 19 °C). performed in this study; the other ones are data of Arzilli and Carroll Decompression + cooling experiments: d D17 (Pf =150MPa, (2013)

from glass was difficult using only thresholding segmentation The batch nucleation rate, Im (Hammer and Rutherford and many crystals had to be defined manually. The areas of 2002;Couch2003; Brugger and Hammer 2010a;Sheaand clinopyroxenes, oxides, and bubbles were measured using Hammer 2013) was calculated from the following: simple thresholding because their gray tones were clearly dis- N v tinguishable. The uncertainty in the area measurements was I m ¼ ð5Þ estimated on the basis of 10 images per sample. An area from t 2 0.022 to 0.58 mm was used for image analyses of natural where t is the experimental duration. Alkali feldspar crystal samples. dimensions were also measured on BSE images using ImageJ For the synthetic samples, the following parameters were software. Typically, only the largest 10 crystals in each image also calculated. Mean crystal size (sn) was also calculated were measured (e.g., Couch 2003)todeterminethemaximum through the following equation: growth rate. Growth rate (YL) was calculated using only the rffiffiffiffiffiffi ϕ longest dimension of each crystal (e.g., Fenn 1977; Swanson Sn ¼ ð3Þ 1977; Hammer and Rutherford 2002; Couch et al. 2003). The N a uncertainty for sizes and growth rate measurements was esti- mated on the basis of the 10 largest crystals observed in each The volumetric number density (Nv) calculation uses a standard method for correcting the area to volumetric nucle- sample. The maximum growth rate of alkali feldspar was cal- ation density (Cheng and Lemlich 1983;Couch2003): culated through the following relation (Swanson 1977; Couch et al. 2003; Shea and Hammer 2013): : ¼ N a ð Þ 0 5L N v 4 Y L ¼ ð6Þ Sn t Bull Volcanol (2016) 78: 72 Page 9 of 20 72 where t is the duration of the experiment. The mean growth Geofisica e Vulcanologia, sezione Osservatorio Vesuviano rate (Y Sn ) was calculated from the equation (Brugger and (Naples, Italy), as described in Mormone et al. (2014). Hammer 2010a, b): Operating conditions were CuKα radiation, 40 kV, 40 mA, 2θ range from 3° to 100°, equivalent step size 0.0179° 2θ, ¼ Sn ð Þ equivalent counting time 298.09 s per step. The X’Pert High Y sn 7 t Score Plus 2.2 d software allows qualitative and semi- quantitative data of the analyzed natural powders. Semi- The difference between YL and Y Sn is that YL is the maxi- mum growth rate obtained during the experiment, whereas quantitative XRD analysis was carried out on 7 samples using the X’Pert High Score Plus 3.0e package interface. The X-ray Y S is the mean growth rate calculated using the mean crystal n diffraction analysis, without internal standard, can provide size (Sn). semi-quantitative phase proportions which are reported on an amorphous-material-free basis. Detection limit for phase Crystal size distribution analysis abundance is ∼0.3 wt%.

Fragments from selected natural samples were used for CSD measurements to estimate the magma residence time based on Results the relationship between feldspar population density, crystal sizes, and experimentally determined growth rates of alkali Experimental samples: crystallization kinetics feldspar. CSD studies provide quantitative information on re- and textural features lations between crystal population density and crystal length for a population of crystals. The linear relation provides esti- The experimental run products contain one or more of the mates of timescales of magmatic processes (Marsh 1988; following phases: alkali feldspar, clinopyroxene, Fe-Ti oxide, Cashman and Marsh 1988). The slope of the correlation is biotite, sodalite, glass and vesicles (Table 3). Sodalite − equal to 1/(growth rate × residence time), and the intercept (Table 3) is formed only at pressures ≤50 MPa, and it occurs is equal to the nucleation density (Higgins 2000). in experiments with high crystal fractions of alkali feldspar Crystal dimensions and abundances of each size population (ϕ). The abundances of clinopyroxene, biotite, and magnetite were recovered through image analyses of BSE images and remain low (0.01 to 0.10) for the trachytic bulk composition. Adobe Photoshop ® software. The crystals were manually Few quench feldspar crystals are present in our samples. segmented because of the weak contrast between trachytic Some tiny, acicular, and curved-shape feldspar crystals are glass and feldspar microlites. We were able to define the con- visible in the run products (Fig. 5). They were not included tour of each single crystal also in agglomerate structures. in the data that were used to estimate the kinetics of crystalli- However, for highly crystalline samples, defining the geome- zation. Because we used cold seal pressure vessels fitted with try of each crystal involved some interpretation. Several as- rapid quench extensions, the quench was very rapid and sumptions were made to define different crystal geometries in quench crystal formation was minimal. agglomerates but they provided similar results. The recon- The experiments were texturally analyzed to quantify the structed image was processed through ImageJ software (NIH number densities (Na), nucleation rates (Im), crystallinities (ϕ), Image; Abramoff et al. 2004; Schneider et al. 2012)toquan- crystal lengths (L) and growth rates (YL) of alkali feldspar tify data on crystallinity percentage and crystal sizes. The (Table 3). The experiments between 50 and 200 MPa pressure relationship between crystal population density and crystal show that the number density of alkali feldspar ranges from length for a population of crystals was then obtained using 4 6 −2 10 and 10 cm (Table 3) and the order of magnitude of Na the CSD Corrections 1.3 program (Higgins 2000, 2002). increases with decreasing pressure. The run products have a wide range of crystallinities of alkali feldspar (ϕ)thatvaries X-ray diffraction analysis between 0.01 and 0.95. The cooling experiments produced ϕ ranges from 0.01 to 0.93, whereas decompression + cooling Selected whole-rock samples and related hand-picked shards experiments exhibit ϕ variations from 0.05 to 0.72. The feld- from the same sampled units used for CSD were examined by spar fraction ϕ reaches a maximum of 0.95 at 30 MPa and X-ray powder diffraction (XRD). Samples were powdered in long duration (14 h) during an isothermal decompression an agate mortar and analyzed (∼1 g per sample) by XRD to experiment. further characterize mineralogy and to make semi-quantitative The experiments at high pressure (100, 150, and 200 MPa) estimates of phase proportions on a larger amount of sample and short experimental duration (7200–28,800 s) show that ϕ than that studied in thin sections. changes from 0.01 to 0.32 with increasing duration (Table 3), The XRD patterns were obtained by using a Panalytical whereas for the same pressures but at longer duration X’Pert Pro diffractometer of the Istituto Nazionale di (57,600 s) ϕ is ∼0.42. For the experiments at 50 MPa and 72 Page 10 of 20 Bull Volcanol (2016) 78: 72

Table 3 Experimental results of nucleation and growth of alkali feldspar during isothermal cooling, decompression + cooling, and isothermal decompression experiments

−2 −3 −1 Sample Na (cm ) ϕ Im (cm s ) L (cm) YL (cm/s) Phases Y Sn (cm/s)

Isobaric cooling experiments D20 4.6E + 04(1) 0.32(0.02) 6.2E + 02(3) 0.0260(0.0095) 4.5E−07(9) 9.0E−08(3) gl-af-cpx-mt-bt-bl D15 nd 0.01 nd nd nd nd gl-af-cpx-mt-bt-bl D13 1.31E + 04(1) 0.04(0.01) 2.6E + 02(3) 0.0116(0.0020) 2.0E−07(4) 6.1E−08(8) gl-af-cpx-mt-bt-bl D6 4.62E + 04(1) 0.17(0.01) 1.11E + 03(3) 0.0182(0.0032) 2.8E−07(4) 8.9E−08(3) gl-af-cpx-mt-bl D11 2.90E + 05(4) 0.29(0.01) 4.0E + 04(1) 0.0107(0.0035) 7.4E−07(2) 1.38E−07(3) gl-af-cpx-mt-bl D12 1.26E + 06(1) 0.32(0.01) 1.74E + 05(3) 0.0096(0.0019) 3.3E−07(7) 3.49E−08(4) gl-af-cpx-mt-bl D33 7.1E + 06(1) 0.49(0.01) 1.85E + 06(4) 0.0041(0.0006) 1.4E−07(2) 1.84E−08(3) af-cpx-mt-gl-bl D10 6.97E + 05(2) 0.52(0.02) 8.9E + 02(4) 0.0116(0.0020) 2.0E−07(3) 9.4E−08(2) af-cpx-mt-gl-bl D9 1.41E + 06(2) 0.55(0.01) 3.92E + 04(8) 0.0045(0.0011) 3.9E−08(9) 1.09E−08(2) af-cpx-mt-gl-bl D69 nd 0.93(0.02) nd nd nd nd af-cpx-mt-gl-bl-sod D78 nd 0.93(0.02) nd nd nd nd af-cpx-mt-gl-bl-sod Decompression + cooling experiments D17 1.23E + 04(1) 0.05(0.01) 2.8E + 02(3) 0.0092(0.0022) 2.13E−07(8) 9.3E−08(9) gl-af-cpx-mt-bt-bl D25 7.11E + 04(6) 0.41(0.04) 5.1E + 02(3) 0.0174(0.0032) 1.51E−07(3) 4.2E−08(2) gl-af-cpx-mt-bt-bl D29 4.92E + 04(5) 0.42(0.01) 2.92E + 02(5) 0.0185(0.003) 1.6E−07(3) 5.07E−08(7) gl-af-cpx-mt-bl D21 9.82E + 06(3) 0.72(0.03) 6.3E + 05(1) 0.004(0.0012) 3.5E + 08(2) 4.7E−09(1) af-cpx-mt-gl-bl Isothermal decompression experiments D62 nd 0.95(0.01) nd nd nd nd af-cpx-mt-gl-bl-sod

Glass and crystalline phases are listed in approximate order of decreasing relative abundances ϕ Na nucleation density, volume fraction, Im nucleation rate, L maximum length, YL maximum growth rate, Y Sn mean growth rate. The value in parentheses is the standard deviation of the mean value. nd not determined phases presents because the contrast between several phases is not sufficient to distinguish them in BSE images. gl glass, af alkali feldspar, cpx clinopyroxene, mt magnetite, bt biotite, bl bubbles

15 ≤ΔT ≤ 40 °C, ϕ ranges between 0.29 and 0.55 (Table 3). (Brugger and Hammer 2010a). As reported in Table 3,values

Several samples (see D69, D78, D21, and D62 in Table 3) of YL and Y Sn have similar orders of magnitude, but, as ex- – – show that low pressure (30 50 MPa), H2Ocontent(1.7 pected, maximum YL are higher than the mean Y Sn (Fig. 6). 2.7 wt% H2O), and 40 ≤ΔT ≤ 115 °C result in extensive The batch nucleation rates (Im) of alkali feldspar range be- 2 6 −3 −1 crystallization of alkali feldspar (0.72 ≤ ϕ ≤ 0.95). These sam- tween 10 and 10 cm s , showing an increase at low Pf ples have strongly intergrown textures characterized by a large and high ΔT (Table 3). quantity of tiny crystals (tabular and acicular crystal inter- growth) with very low abundance of glass (Fig. 4a–i). Alkali Natural samples: textural, XRD, and CSD data feldspars commonly have tabular and elongated rectangular shapes (Figs. 4a–i and 5b–d) at the different experimental In natural samples, the dominant alkali-feldspar mineral phase conditions. In contrast, spherulitic alkali feldspar crystals are is present in various percentages, from low abundance (nearly more common at higher Pf (= PH2O ), and they are present in glassy matrix; Fig. 3e) to very high abundance (Fig. 3b). XRD both cooling and decompression experiments (Fig. 5). data were used to evaluate the relative proportions of Although it was difficult to obtain reliable chemical analysis, anorthoclase and sanidine and the accessory minerals present it was possible to establish that (1) alkali feldspars have less in the various samples (Table 4). The anorthoclase abundance than 8 mol% of An and 39 to 71 mol% of Or and (2) residual is higher in the highly crystalline brown pumices (71–83 %) glasses are trachytic in composition. and scoriae (55–60 %). These samples also show small

The maximum growth rates (YL) of alkali feldspar range amounts of sodalite (∼3and∼6 %, respectively). XRD data between 10−8 and 10−7cm/s (Table 3). Values at the lower end show trace amounts of biotite (≤0.6 %) in several pumice of this range occur in experiments with longer duration samples (Table 4), whereas it is absent in the scoriae erupted (57,600 s), probably because of the closer approach to equi- at the end of the eruption; these results are also consistent with librium at longer times. Table 3 also reports mean growth rates petrographic observations. Particularly, the basal yellowish ∼ (Y Sn ) determined using the batch method as described above and the brown pumices have 0.5 wt.% of biotite. A Bull Volcanol (2016) 78: 72 Page 11 of 20 72

and another, with shallower slope, for larger crystals; trends for different samples are nearly parallel. In comparison, CSDs in the literature (Piochi et al. 2005) show comparable intercept values but smaller crystal sizes as a consequence of using smaller sample areas for CSD measurements; therefore, this should result in under-estimation of the less abundant larger crystals. The stratigraphically lowermost pumices show a sim- ple CSD with a unique growth trend, whereas scoriae are characterized by two trends that suggest two distinct periods of crystal growth (Fig. 7). As a whole, the smaller microlites define CSD curves with slopes that range from −22 to −8and an intercept at 11 to 19 mm−4, whereas the curves for larger microlites have slopes between −2and−7 and extrapolated intercepts at 9–11 mm−4 (Table 5). Lower slopes are generally determined for the lower vesicular brownish bands and black scoria (c2s 9-12–1 and MN4 top 06–1, respectively, in Table 5), whereas higher slopes are from the highly vesicular yellow glass. Fig. 6 Comparison between growth rate values (YL and Y Sn )determined by different methods (see BSEM and image analyses^ section). YL is based on maximum feldspar size (see Eq. 6 in BSEM and image ^ analyses section), whereas, the Y Sn is derived by total number and area of feldspar crystals (see Eq. 7 in BSEM and image analyses^ Discussion section) following Brugger and Hammer (2010b) TheroleofpressureandΔT in changing crystallinity comparison of XRD semi-quantitative data on whole-rock and related separated shards from each sample indicate compara- By combining our results with those of Arzilli and Carroll ble mineral phase percentages. This shows that the contribu- (2013), we show that isobaric cooling experiments at 50 and tion of phenocrysts is not significant, and XRD data can there- 70 MPa can produce higher ϕ than those obtained at 150 and fore be used to assess microlite abundance in combination 200 MPa (Fig. 8a, b). Furthermore, the relation between Na Δ with BSE images. and T as a function of Pf for cooling experiments (Fig. 8c) Δ Generally, pumices are banded, but the individual bands shows that at 70 and 100 MPa, Na increases as T increases, show no significant contortions (∼linear). Alkali feldspars whereas for experiments at 50 MPa, Na is more constant with Δ are aligned within bands of pumices and, sometimes, clustered T. The diagram in Fig. 8d shows also that Na is almost con- Δ in spherulitic clots. The size and abundance of alkali feldspar stant with time. Our results show that at low T, Na increases determined in natural samples allowed us to obtain measure- as Pf (= PH2O ) decreases (Fig. 8d), and this indicates that the ments of CSDs for several samples (Fig. 7). In agreement with nucleation process is favored at low pressure, with lower melt previous analysis (D’Oriano et al. 2005; Piochi et al. 2005; H2O contents. Mastrolorenzo and Pappalardo 2006), the size distributions of In isothermal decompression and decompression + cooling microlites can be described by two approximately linear experiments, the high crystallinity textures are produced at 30 trends, a steeper one for crystals smaller than about 0.6 mm, and 50 MPa (Fig. 9a, b). Figure 9c shows that Na for experi- ments at 30 and 50 MPa (∼106.5cm−2) is 1.5 log unit higher ∼ 5 −2 Table 4 Mineral phase abundances (wt%) in the Monte Nuovo than experiments at 70, 100, and 150 MPa ( 10 cm )for products as determined by semi-quantitative analyses of the XRD data ΔT = 40 °C. All experiments have about the same Na (∼105cm−2)atΔT <40°C(Fig.9c). Probably, small ΔT could Sample Anorthoclase Sanidine Sodalite Biotite Magnetite slow down the nucleation process, favoring crystal growth. As LM C3 wr 70.9 24.8 3.4 0.6 0.4 with the isobaric cooling results, nucleation of alkali feldspar LM C3 sh 82.6 14.2 2.8 0.2 0.3 appears to be strongly favored at lower Pf (Fig. 9d). LM inf l wr 35.9 63.7 – 0.5 – The isothermal decompression experiments show the effect LM inf l sh 34.9 64.9 – 0.2 – of the undercooling (effective undercooling) and pressure (Pf) LM inf d sh 39.8 59.6 – 0.5 0.1 on the crystallization kinetics of alkali feldspar (Fig. 10a, b), MN4 wr 59.5 33.6 5.8 – 1.1 allowing us to better understand the role of these parameters MN4 sh 54.9 39.2 5.6 – 0.3 during magma ascent. Time-temperature-transformation (TTT) diagrams (Fig. 10a) are useful for showing the effect Goodness of fit (GOF) <3; refer to text for analytical procedures of undercooling and viscosity on transformation kinetics 72 Page 12 of 20 Bull Volcanol (2016) 78: 72

Fig. 7 CSDs of early erupted low-crystalline and highly vesicular pumice (left), brown pumices intermediate in the stratigraphic sequence (central diagram) and scoriae (right diagram). Note the occurrence of two microlite populations and the similarity of slopes for smaller microlites

(Kouchi et al. 1986;Putnis1992; Porter and Easterling 1997; (Fig. 10a). This result can be important for quantifying and Hammer and Rutherford 2002; Chevrel et al. 2015). The TTT modeling crystallization of trachytic magmas in response to plot for the crystal fraction variations was determined from the magma degassing during ascent in the conduit. Magmas en- results of the isothermal decompression experiments as a tering shallow volcanic conduits usually contain crystals, and function of ΔT, Pf and time (Fig. 10a). The crystal fraction transformation kinetics might be dominated by either feldspar (ϕ) increases as ΔT and time increase and Pf decreases nucleation or growth depending on ΔT-time paths, potentially

Table 5 Crystal growth Sample Slope Intercept (mm−4) t (h)–mean t (h)–min t (h)–max t (days)–max rates (YL) as derived by r r r r experiments and related residence − time (tr) infl 8-12 20.8 12.8 6.1 1.1 121.6 5.1 infb 01-10 −21.6 15.7 5.8 1.1 117.2 4.9 infl 1-7 −12.3 13.2 10.2 1.9 204.8 8.5 infb 1-4 −15.6 16.2 8.0 1.5 161.7 6.7 infsc 02-05 −18.5 16.0 6.8 1.3 136.4 5.7 infb ch07-10 −16.1 13.3 7.8 1.4 156.5 6.5 c1 1-7 −9.7 11.3 13.0 2.4 261.1 10.9 c1 8-13 −17.0 15.8 7.4 1.4 148.9 6.2 c2c 4-7–1 −2.3 9.3 54.7 10.1 1099.5 45.8 c2s 9-12–1 −2.3 10.0 55.6 10.2 1117.4 46.6 c2c 4-7–2 −8.1 16.8 15.6 2.9 313.2 13.1 c2s 9-12–2 −13.7 17.3 9.2 1.7 184.4 7.7 MN4 top 06-1 −6.7 11.0 18.9 3.5 378.8 15.8 MN4 top 06–2 −12.9 14.7 9.7 1.8 195.1 8.1 c3 1-2 −19.6 16.3 6.4 1.2 128.6 5.4 c3 1-2 −20.4 17.5 6.2 1.1 123.7 5.2 MN4top 02-05 −20.0 19.0 6.3 1.2 126.3 5.3

Experimental growth rate mean YL = max YL = min YL = cm/s 2.21E−07 1.20E−06 1.10E−08 mm/h 7.96E−03 4.32E−02 3.96E−04

Crystal growth rates (YL) are average of the mean, minimum, and maximum values obtained considering YL obtained in this study and from Arzilli and Carroll (2013). Residence times (tr) are calculated through the equation: tr =(−1/YL × slope). Slopes and intercepts as derived by CSD Corrections 1.3 program (Higgins 2000, 2002). CSD diagrams shown in Fig. 7. tr mean, tr min, and tr max are the mean, minimum, and maximum residence times, respectively. Growth rates obtained from experiments are italicized Bull Volcanol (2016) 78: 72 Page 13 of 20 72

Fig. 8 Textural evolution of alkali feldspar as a function of ΔT, Pf and texp texp. c Na as a function of ΔT. d Na as a function of texp. Diagrams show in cooling experiments. a Crystal fraction of alkali feldspar (ϕ)asa data obtained in this study and from the previous experimental work of function of ΔT. b Crystal fraction of alkali feldspar (ϕ)asafunctionof Arzilli and Carroll (2013) resulting in large, rapid changes in magma rheology. trachytic melts. The comparison between isobaric

Figure 10b shows the influence of ΔT and Pf on nucleation cooling, isothermal decompression, and decompression + and growth rates at constant time (1 h), and the results dem- cooling experiments shows that similar conditions, in onstrate that crystal growth is promoted at relatively low ΔT terms of Pf, ΔT, and time, may produce similar Na, ϕ, and high pressure, whereas nucleation rates increase with in- Im and YL. This suggests that it is difficult to distinguish creasing ΔT and decreasing pressure (Fig. 10b). between conventional undercooling (ΔT, related to iso-

Our results combined with those of Arzilli and Carroll baric cooling) and effective undercooling (ΔTeff, related (2013) show that pressure, ΔT, and time can play a fun- to isothermal decompression) when a single step of damental role in controlling the crystallization kinetics of cooling and decompression occurs. 72 Page 14 of 20 Bull Volcanol (2016) 78: 72

Fig. 9 Textural evolution of alkali feldspar as a function of ΔT, Pf,and afunctionofΔT. d Na as a function of texp. Diagrams show data obtained texp in decompression + cooling and isothermal decompression in this study and from the previous experimental work of Arzilli and experiments. a Crystal fraction of alkali feldspar (ϕ) as a function of Carroll (2013) ΔT. b Crystal fraction of alkali feldspar (ϕ)asafunctionoftexp. c Na as

The separate effects of pressure and ΔT on crystalli- melt H2O content on nucleation process, as observed by zation of alkali feldspar are particularly difficult to dis- Hammer (2004) for rhyolitic melts and by Arzilli et al. tinguish. The main result provided by our combination (2015) for high-K basalts. Furthermore, our results show of isobaric cooling, isothermal decompression, and de- that the increase of crystallinity and Na can occur in a compression + cooling experiments is that the nucle- few hours (Figs. 8b, d, 9b, d, and 10a). This implies ation of large numbers of crystals is favored at low Pf that the rheology of rising trachytic magma could (Figs. 4, 8,and9) and not necessarily only at high ΔT. change quickly, affecting the eruptive style.

In fact, for certain conditions of low Pf and low ΔT (Fig. 9d), nucleation was strongly favored. Our results Magma residence time deduced from growth rates show that nucleation is reduced at high pressure (Figs, and CSDs 5, 8, 9,and10b), demonstrating that a dramatic increase of crystal number density can be induced at low pres- Measurements of CSD, when combined with experimen- sure in trachytic melts, likely as a result of the effect of tal growth rate data for alkali feldspar (YL), can help us Bull Volcanol (2016) 78: 72 Page 15 of 20 72

deposit of Monte Nuovo. Furthermore, size- proportionate growth of crystals, which has been largely observed and modeled in low-viscosity aqueous solu- tions (Eberl et al. 1998, 2002;KileandEberl2003), must be excluded. Therefore, the considerations above support the assumption of (approximately) constant growth rate and variable nucleation rate through time in viscous magmas. In addition, the textural similarities

(in both Na and ϕ) between synthetic and natural sam- ples (Fig. 11) could indicate the main role of pressure on nucleation rate and decompression of magma as the dominant processes during the Monte Nuovo eruption.

Theincreaseofϕ and Na through experimental time (Figs. 9band10a) could definitively affect the CSDs. However, the concave-up CSD patterns could be an ar- tifact caused by the presence of two populations, in line with previous works (Piochi et al. 2005 and D’Oriano et al. 2005) and experiments (Brugger and Hammer 2010a). We use new data on both CSDs (Fig. 7)andfeldspar growth rates to make new estimates of crystallization times

(tr) for the Monte Nuovo magma (Table 5). The estimated −8 growth rates (YL) mostly vary between 1.1 × 10 and 7.2 × 10−7cm/s, with one higher value at 1.2 × 10−6cm/s, −7 and the mean YL equal to 2.21 × 10 cm/s (Table 5). Since

YL is slightly higher than Y Sn (Fig. 6), we used the former as it is the maximum growth rate, and therefore, it allows us to estimate the lower limit on the timescale of the magma resi- dence. The variation of magma residence time (Table 5)is related to the range of growth rates and CSD slopes, and estimated crystal growth timescales range from few months

to few hours. The mean YL (here the mean YL is considered the most representative growth rate) and the CSD slopes suggest that magma was stored for a few days in the subsurface (ob- Fig. 10 Crystallization kinetics of isothermal decompression tained from the flatter CSD slopes) and the ascent up to experiments. a Time–temperature–transformation (TTT) diagram constructed using the crystal fraction (ϕ) variation as a function of ΔT quenching lasted a few hours (obtained from the steep CSD and time. b The variation of nucleation rate (Im) and growth rate (YL)with slopes) (see Fig. 7 and Table 5). These residence times appear ΔT realistic when compared with timing of phenomena observed prior to eruption (see BThe Monte Nuovo eruption^ section; Fig. 1). Notably, the steep CSD slopes (i.e., correspond to to better constrain the time of magma crystallization short residence times) characterize the early erupted sodalite- (e.g., Marsh 1988; Cashman and Marsh 1988; Higgins free pumices, indicating the abrupt quenching of magma ris- 1996;Piochietal.2005;D’Oriano et al. 2005;Armienti ing from the deeper levels. The flatter CSD slopes (i.e., corre- et al. 2007; Brugger and Hammer 2010a). The empirical spond to long residence times) characterize the most crystal- approach is largely used by geologists and engineers, line matrices, in which the anorthoclase is more abundant than although different assumptions must be met (Eberl sanidine in scoriae and brown pumices (Table 4). et al. 1998, 2002; Brugger and Hammer 2010a). A batch crystallization mechanism in which no mass was Implication for magma dynamics during the Monte added or subtracted to the crystallizing system is re- Nuovo eruption quired (e.g., Marsh 1988). This assumption is satisfied by the homogeneous mineral assemblage, the proportion Experimental alkali feldspars have elongated rectangular of phases, and the chemical composition of natural crystal shapes, and they occur in clots (Fig. 4)oriso- products (Piochi et al. 2005, 2008), related to each latedintheglass(Fig.5), similar to their occurrence in 72 Page 16 of 20 Bull Volcanol (2016) 78: 72

Fig. 11 Textural evolution of alkali feldspar as a function of ΔT and Pf. Na and ϕ of cooling and decompression experiments were obtained from Textural evolution of alkali feldspar in isobaric cooling (a), (c) and, in batch calculation, whereas Na and ϕ of natural samples of Monte Nuovo isothermal decompression and decompression + cooling (b), (d) were obtained from CSD results. Diagrams show data obtained in this experiments, compared with that in natural samples of Monte Nuovo study and from the previous experimental work of Arzilli and Carroll (pumices and scoriae). a, b The nucleation density (Na)andthecrystal (2013) fraction (ϕ) of alkali feldspar were obtained from batch calculation. c, d the natural pumices (Fig. 3). Both isothermal decom- produced material texturally comparable with the natural pression and decompression + cooling experiments at products, although the pyroxene occurrence, albeit not 50 MPa produced textures similar to those of natural abundant (<0.10), represents the most important discrep- scoriae, with high crystallinity and aligned tabular or ancy. This may depend on the starting experimental elongated rectangular crystals (see Figs. 3band5c), conditions because pyroxenes are present in the ZAC together with magnetite and sodalite crystals (Table 4). starting material, and the oxygen fugacity of the CSPV The highly crystalline samples also show vesicles with (∼NNO + 1 log units) may also be different from that irregular margins similar to the natural scoriae. of the Monte Nuovo system. Furthermore, the Or content of the synthetic alkali feld- Decompression (isothermal decompression and decom- spars and those of natural ones is similar, ranging be- pression + cooling) experiments show two populations with tween40and70mol%(D’Oriano et al. 2005;Piochi different nucleation density (Fig. 11b, d). The first population et al. 2005;Piochietal.2008). The experiments at low pressure (30 MPa < Pf < 50 MPa) has values of Na Bull Volcanol (2016) 78: 72 Page 17 of 20 72 between 106 and 107cm−2, exceeding values observed in the natural samples (Fig. 11b, d). The second group of experi- ments at high pressures (100 MPa < Pf < 150 MPa) has values 4 5 −2 of Na between 10 and 10 cm , similar to the poorly crystal- lized pumice (Fig. 11b, d). Instead, ϕ ranges from ∼0.10 to ∼0.60, showing values similar to and higher than those of natural pumices (Fig. 11b, d). The experiments at 50 MPa (sample D81), 70 MPa (sample D82), 100 MPa (sample D80), and 150 MPa (sample D55) are characterized by 5 −2 Na = ∼10 cm and ϕ between ∼0.10 and ∼0.30 (Arzilli and Carroll 2013), therefore comparable with poorly crystallized natural pumice (Fig. 11b, d). Isothermal decompression exper- iments show that the crystallinity of natural samples could increase during magma ascent, because the increase of ΔT and time and the decrease of pressure produce an increase of ϕ (Fig. 10a). Furthermore, the increasing number of crystals with small size in scoriae (Fig. 7) could be related to the increase of Im with ΔT (low pressure) (see Fig. 10b). In fact, the comparison between experimental nucleation rates and those obtained from CSD suggests that the main crystalliza- tion of the scoriae occurred from 50 to 30 MPa (Fig. 12a). The ϕ of alkali feldspar = 0.01–0.30 in the pumices implies ΔT <50°CandP > 70 MPa; the increase in ϕ up to 0.40– 0.60 in the scoriae, as well as the highest one at 0.95 (D’Oriano et al. 2005), requires higher ΔT and low pressure. The experimental results suggest that the crystallization of alkali feldspars in magma erupted during the Monte Nuovo eruption started between <200 and 100 MPa when temperature was close to the liquidus of alkali feldspar (825–840 °C) (see Fig. 1 in Arzilli and Carroll 2013). This is also supported by the low amount of biotite in natural products (Table 4); in fact, previous experimental work and the present results (see samples D20, D15, D17, D25 in Tables 1 and 2) on trachytic melts indicate the stability field of biotite at pressure >135 MPa (Fabbrizio and Carroll 2008). Based on these results, we constrain the pre-eruptive magma storage at a depth consis- tent with biotite crystallization (>4 km), slightly deeper than hypothesized by Piochi et al. (2005). Furthermore, Fig. 12 Comparison between experimental nucleation rates (Im) and those obtained from CSD results. a The diagram reports Im obtained our experimental data indicate that the early erupted from cooling experiments and from CSD results of pumices and scoriae – Monte Nuovo magma (LM unit) stayed at 30 50 MPa of Monte Nuovo (MN). b) The diagram reports Im obtained from (∼2 km of depth) for only a short time. Sodalite absence isothermal decompression and decompression + cooling experiments and biotite occurrence in these samples support such baro- and from CSD results of pumices and scoriae of Monte Nuovo (MN) metric conditions. Low ΔT can be suggested for the early erupted products, supporting previous inferences (D’Oriano crystallinity and the enrichment of Cl in the residual melt et al. 2005). Our results indicate lower pressures for the that occurs at low pressure (Carroll 2005). This is in agree- later erupted magma (i.e., scoriae in the Upper Member) ment with the content of Cl in brown pumices and scoriae, registers. The high groundmass crystallinity with alkali which is relatively higher than that of other pumices feldspar as the dominant phase and the sodalite occurrence (Piochi et al. 2008). Decompression was the main process in these later erupted fragments (Table 4) suggest effective controlling the textures of the Monte Nuovo products, in undercoolings of several hundreds degrees Celsius and agreement with occurrence of microlites that are curved pressure of crystallization below 50 MPa (see Tables 2 following the morphology of adjacent vesicles, suggesting and 3). The sodalite occurrence should be due to the high that they grew concomitantly with vesicles (Piochi et al. 72 Page 18 of 20 Bull Volcanol (2016) 78: 72

2005, 2008). In particular, our results support the idea that conduit (the crystallization in the brown pumice and scoriae decompression and a decrease in H2O solubility promoted occurred over a longer time compared with pumice sample extensive crystallization in later erupted magma (Fig. 3), crystallization), as shown from our results (see samples D69, moving the melt towards the most evolved trachyte com- D78, and D62). Increased ΔT can enhance crystal formation positions observed (D’Oriano et al. 2005;Piochietal. by increasing crystal nucleation rate (Fig. 10), changing the 2008). magma rheology toward more viscous behavior as crystal Magma ascent from the storage depths (≤135 MPa) to content increases. Furthermore, Melnik (2000) shows that shallower depths (≤50 MPa) occurred rapidly, with timescales the high viscosity of the magma in the conduit is probably on the order of hours suggested by our CSD results (see responsible for increasing pressurization, and our results jus- Table 5). Fabbrizio and Carroll (2008) also suggested that tify a potential overpressure in the conduit, which may have the travel of the biotite-bearing phlegraean magma from res- triggered the final vulcanian explosion of Monte Nuovo. ervoir at depth of c. 7 km (∼200 MPa) to the surface was rapid, Finally, the coexistence of pumices with different vesicle based on dissolution rate data of biotites (hours to a maximum and microlite textures (for example, samples LM inf l, LM in of 1–2 days; Fabbrizio and Carroll 2008). Low ascent rate d, and LM c3) suggests a complex process of magma ascent. results in dissolution of biotite. Therefore, the magma ascent Mingling between magmas that reached the surface through rate is estimated in 400–100 m/h from a depth of 4–5km different pathways and pressure–temperature–time conditions (biotite-stable). Magma storage of some days in the shallow in the subsurface or lateral variations of magma conditions in crust is indicated by the occurrence of some phenocrysts (that the conduit (D’Oriano et al., 2005) are possibilities to be fur- define CSD slope > −10, Table 5, Fig. 7), whereas the smaller- ther investigated. sized crystals (and the CSD with steeper slopes; Table 5 and Fig. 7) were able to crystallize in a few hours. These results better constrain previous reconstruction of this eruption Conclusions (Piochi et al. 2005;D’Oriano et al. 2005) and suggest the utility of similar studies for assessment of volcanic hazards. The comparison between experimental data and natural data

The detected CSD variation reflects the textural features of provides a set of P–T–H2O content conditions for trachytic samples, resulting from small differences in rates of magma magmas valid for the Monte Nuovo eruption at Campi ascent, crystallization, and degassing (see also D’Oriano et al. Flegrei. The cooling and decompression experiments 2005;Piochietal.2005, 2008) during eruption. The LM ve- reproduced textures observed in pumices and scoriae pro- sicular pumices (early erupted; Fig. 1) register the magmatic duced during the 1538 AD eruption. Our results indicate that fragmentation generated by magma-water interaction follow- the juvenile pyroclastic components provide a record of mag- ing gas expansion, while syn-eruptive outgassing-induced ma ascent conditions within the volcanic conduit. This erup- crystallization produced brown pumices and scoriae (later tion requires a water-saturated magma stored at around 150– erupted magma; Fig. 1) in the course of the eruption. The most 135 MPa, near the liquidus temperature of alkali feldspar; its crystalline and less vesicular scoriae contain a lower amount ascent toward the surface lasted a few hours to several days. of H2O than LM microlite-poor and vesicle-rich pumices Our results do not furnish direct information concerning what (Piochi et al. 2008). triggered the eruption at Monte Nuovo. The phlegraean trachyte can rapidly crystallize large crystal As a whole, this experimental study offers a set of pressure, fractions of alkali feldspar, over small pressure-temperature temperature, and undercooling (water-saturated) conditions, intervals (Figs. 10, 11,and12), and this can strongly modify useful to constrain the timescale and to better understand the the rheology of a trachytic magma, favoring rapid changes in magma behavior of numerous phlegraean eruptions, using the eruptive style or ending of the eruption. Our results allow us to textures of their trachy-phonolitic products. In fact, alkali feld- constrain the processes producing the change in eruptive style spar is the main crystal phase in the phlegraean rocks and, as from pumice to scoria type deposits due to an increase of shown for the Monte Nuovo eruption, its crystallization could magma viscosity (Caricchi et al. 2008; Vona et al. 2011), be occurred between 4 and 7 km depth (∼100–200 MPa) and quantitatively supporting the idea of D’Oriano et al. (2005) proceed at shallow depths, where the magma ascent through about plug generation in the volcanic conduit in the last erup- the conduit was faster and texture was effectually frozen in tive phase. In fact, values of ϕ between 0.80 and 0.90 ob- (mostly between 2 and 3 km; 50–70 MPa). The absence of served in the natural scoriae of Monte Nuovo produced in this sodalite in most of the natural samples implies a rapid magma phase requires higher ΔT and lower P, as suggested from the ascent at 30–50 MPa (pressure for sodalite stability) or experiments. The abundance of crystals in the scoria deposits magmas stored at these levels only occasionally were erupted. reaching values higher than 90 % (Fig. 11a, b; D’Oriano et al. Furthermore, the temperature of the magma, in pre-eruptive

2005) may be related to high ΔT and low Pf or longer resi- conditions, is most likely between 825 and 840 °C (close to dence of the magma in a shallower part (Pf <50MPa)ofthe the liquidus of alkali feldspar). Bull Volcanol (2016) 78: 72 Page 19 of 20 72

Acknowledgments This work was supported by the 2005–2006 Cashman KV, Marsh BD (1988) Crystal size distribution (CSD) in rocks INGV-DPC project V3-2/UR14 (M. Piochi) and PRIN 2009 (R. and the kinetics and dynamics of crystallization II: Makaopuhi lava Moretti), FAR 2012, and PRIN 2009 (M. R. Carroll). We warmly thank lake. Contrib Mineral Petrol 99:292–305 the Associate Editor P. J. Wallace for careful handling of the manuscript Cheng HC, Lemlich R (1983) Errors in the measurement of bubble size and for useful corrections and comments that have substantially improved distribution in foam. Ind Eng Chem Fundam 22:105–109 it. The authors acknowledge C. Martel and an anonymous reviewer for Chevrel MO, Cimarelli C, deBiasi C, Hanson JB, Lavallee Y, Arzilli F, the detailed and constructive comments that greatly helped the authors in Dingwell DB (2015) Viscosity measurements of crystallizing andes- improving the manuscript. We are grateful to N. Cennamo (II University ite from Tungurahua (Ecuador. Geochem Geophys Geosyst of Napoli, Naples, Italy) for help with the BSE image acquisition and 16:870–889 Luigi Zeni (II University of Napoli, Naples, Italy) that allowed access to Couch S (2003) Experimental investigation of crystallization kinetics in a the laboratory. The VULCAMED project provided funds to install in haplogranite system. Am Miner 88:1471–1485 October 2012 the XRD laboratory at the Istituto Nazionale di Geofisica Couch S, Sparks RSJ, Carroll MR (2003) The kinetics of degassing- e Vulcanologia, in Naples. We would like to thank P. Scarlato, C. Freda, induced crystallization at Soufriere Hills volcano, Montserrat. J and A. Cavallo for assistance with the SEM at INGV, Rome. We warmly Petrol 44:1477–1502 thank M. W. Schmidt for providing SEM access at ETH of Zurich D’Oriano C, Poggianti E, Bertagnini A, Cioni R, Landi P, Polacci M, Rosi (Institute of Geochemistry and Petrology). We also thank P. Landi for M (2005) Changes in eruptive style during the A.D. 1538 Monte SEM analysis and M. R. Cicconi for assistance during experiments. Nuovo eruption (Phlegrean Fields, Italy): the role of syn-eruptive crystallization. Bull Volcanol 67:601–621 Open Access This article is distributed under the terms of the Creative Del Gaudio C, Aquino I, Ricciardi GP, Ricco C, Scandone R (2010) Commons Attribution 4.0 International License (http:// Unrest episodes at Campi Flegrei: a reconstruction of vertical creativecommons.org/licenses/by/4.0/), which permits unrestricted use, ground movements during 1905–2009. J Volcanol Geotherm Res distribution, and reproduction in any medium, provided you give appro- 195:48–56 priate credit to the original author(s) and the source, provide a link to the Di Matteo V, Carroll MR, Behrens H, Vetere F, Brooker R (2004) Water Creative Commons license, and indicate if changes were made. solubility in trachytic melts. Chem Geol 213:187–196 Di Renzo V, Arienzo I, Civetta L, D’Antonio M, Tonarini S, Vito MD, Orsi G (2011) The magmatic feeding system of the Campi Flegrei : architecture and temporal evolution. Chem Geol 281:227– 241 References Di Vito MA, Isaia R, Orsi G, Southon J, de Vita S, D’Antonio M, Pappalardo L, Piochi M (1999) Volcanism and deformation since Abramoff MD, Magalhaes PJ, Ram SJ (2004) Image processing with 12,000 years at the Campi Flegrei caldera (Italy. J Volcanol Geoth Image J. Biophoton Int 11:36–42 Res 91:221–246 Armienti P, Francalanci L, Landi P (2007) Textural effects of steady state Di Vito M, Lirer L, Mastrolorenzo G, Rolandi G (1987) The 1538 Monte behaviour of the Stromboli feeding system. J Volcanol Geoth Res Nuovo eruption (Campi Flegrei, Italy. Bull Volcanol 49:608–615 160:86–98 Dobson PF, Epstein S, Stolper EM (1989) Hydrogen isotope fractionation Arzilli F, Carroll MR (2013) Crystallization kinetics of alkali feldspars in between coexisting vapor and silicate glasses and melts at low pres- cooling and decompression-induced crystallization experiments in sure. Geochim Cosmochim Acta 53:2723–2730 trachytic melt. Contrib Mineral Petrol 166:1011–1027 E berl DD, Drits VA, Srodon J (1998) Deducing growth mechanisms for Arzilli F, Agostini C, Landi P, Fortunati A, Mancini L, Carroll MR (2015) minerals from the shapes of crystal size distributions. Am J Sci 298: – Plagioclase nucleation and growth kinetics in a hydrous basaltic 499 533 melt by decompression experiments. Contrib Mineral Petrol 170: Eberl DD, Kile DE, Drits VA (2002) On geological interpretations of 1–16 crystal size distributions: constant vs. proportionate growth. Am Mineral 87:1235–1241 Boivin P, Bourdier JL, Camus G, de Goer de Herve A, Gourgaud A, Kieffer G, Mergoil JM, Vincent PM, Auby R (1982) Influence Fabbrizio A, Carroll MR (2008) Experimental constraints on the differ- entiation process and pre-eruptive conditions in the magmatic sys- de la Nature des Magmas sur l’Activité Phreatomagmatique: tem of (Naples, Italy. J Volcanol Geoth Res 171: Approche Volcanologique et Thermodynamique. Bull Volcanol 88–102 45:25–39 Fedele L, Scarpati C, Lanphere M, Melluso L, Morra V, Perrotta A, Ricci Brugger CR, Hammer JE (2010a) Crystal size distribution analysis of G (2008) The Breccia Museo formation, Campi Flegrei, southern plagioclase in experimentally decompressed hydrous rhyodacite Italy: geochronology, chemostratigraphy and relationship with the magma. Earth Planet Sci Lett 300:246–254 Campanian Ignimbrite eruption. Bull Volcanol 70:1189–1219 Brugger CR, Hammer JE (2010b) Crystallization kinetics in continuous Fenn PM (1977) The nucleation and growth of alkali feldspars from decompression experiments: implications for interpreting natural hydrous melts. Can Mineral 15:135–161 – magma ascent processes. J Petrol 51:1941 1965 Geschwind C, Rutherford MJ (1995) Crystallization of microlites during Calzolaio M, Arzilli F, Carroll MR (2010) Growth rate of alkali feldspars magma ascent: the fluid mechanics of 1980-1986 eruptions at in decompression-induced crystallization experiments in a trachytic Mount St Helens. Bull Volcanol 57:356–370 – melt of the Phlegraean Fields (Napoli, Italy. Eur J Mineral 22:485 Guidoboni E, Ciuccarelli C (2011) The Campi Flegrei caldera: historical 493 revision and new data on seismic crises, bradyseisms, the Monte Caricchi L, Giordano G, Burlini L, Ulmer P, Romano C (2008) Nuovo eruption and ensuing earthquakes (twelfth century 1582 Rheological properties of magma from the 1538 eruption of AD. Bull Volcanol 73:655–677 Monte Nuovo (Phlegrean Fields, Italy): an experimental study. Hammer JE (2004) Crystal nucleation in hydrous rhyolite: exper- Chem Geol 256:158–171 imental data applied to classical theory. Am Mineral 89: Carroll MR (2005) Chlorine solubility in evolved alkaline magmas. An 1673–1679 Geophys 48:619–631 Hammer JE, Rutherford MJ (2002) An experimental study of the kinetics Carroll MR, Blank JG (1997) The solubility of H2O in phonolitic melts. of decompression-induced crystallization in silicic melt. J Geophys Am Mineral 82:549–556 Res 107:2021 72 Page 20 of 20 Bull Volcanol (2016) 78: 72

Hammer JE, Cashman KV, Hoblitt RP, Newman S (1999) Degassing and Melluso L, Morra V, Annamaria Perrotta A, Scarpati C, Adabbo M microlite crystallization during pre-climactic events of the 1991 (1995) The eruption of the Breccia Museo (Campi Flegrei, Italy): eruption of Mt. Pinatubo, Philippines. Bull Volcanol 60:355–380 fractional crystallization processes in a shallow, zoned magma Higgins MD (1996) Magma dynamics beneath Kameni volcano, Thera chamber and implications for the eruptive dynamics. J Volc Greece, as revealed by crystal size and shape measurements. J Geophys Res 68:325–339 Volcanol Geoth Res 70:37–48 Melnik O (2000) Dynamics of two-phase conduit flow of high-viscosity Higgins MD (2000) Measurements of crystal size distributions. Am gas-saturated magma: large variations of sustained explosive erup- Mineral 85:1105–1116 tion intensity. Bull Volcanol 62:153–170 Higgins MD (2002) Closure in crystal size distributions (CSD), verifica- Mormone A, Piochi M, Bellatreccia F, De Astis G, Moretti R, Ventura tion of CSD calculations, and the significance of CSD fans. Am GD, Cavallo A, Mangiacapra A (2011) A CO2-rich magma source Mineral 87:171–175 beneath the Phlegraean Volcanic District (Southern Italy): evidence – Iezzi G, Mollo S, Ventura G, Cavallo A, Romano C (2008) Experimental from a melt inclusion study. Chem Geol 287:66 80 solidification of anhydrous latitic and trachytic melts at different Mormone A, Piochi M, Troise C, De Natale G (2014) Il laboratorio di ’ cooling rates: the role of nucleation kinetics. Chem Geol 253:91– Diffrattometria a raggi X dell Osservatorio Vesuviano (Istituto 101 Nazionale di Geofisica e Vulcanologia, Napoli): identificazione e Isaia R, Marianelli P, Sbrana A (2009) Caldera unrest prior to intense stima quantitativa delle fasi in campioni polverizzati. Rapporti – volcanism in Campi Flegrei (Italy) at 4.0 ka B.P.: implications for tecnici INGV 279:7 21 ISSN 2039-7941 caldera dynamics and future eruptive scenarios. Geophys Res Lett Parascandola A (1947) I fenomeni bradisismici del Serapeo di Pozzuoli. 36 L21303 Genovese, Napoli Piochi M, Mastrolorenzo G, Pappalardo L (2005) Magma ascent and Kile DE, Eberl DD (2003) On the origin of size-dependent and size- eruptive processes from textural and compositional features of independent crystal growth: influence of advection and diffusion. Monte Nuovo pyroclastic products, Campi Flegrei, Italy. Bull Am Mineral 88:1514–1521 Volcanol 67:663–678 Kouchi A, Tsuchiyama A, Sunagawa I (1986) Effect of stirring on crys- Piochi M, Polacci M, De Astis G, Zanetti A, Mangiacapra A, Vannucci R, tallization kinetics of basalt: texture and element partitioning. Giordano D (2008) Texture and composition of pumices and scoriae Contrib Mineral Petrol 9:429–438 from the Campi Flegrei caldera (Italy): implications on the dynamics La Spina G, Burton M, de’ Michieli Vitturi M (2015) Temperature evo- of explosive eruptions. Geochem Geophys Geosyst 9 Q03013 lution during magma ascent in basaltic effusive eruptions: A numer- Porter D, Easterling K (1997) Phase transformations in metals and alloys . ical application to Stromboli volcano. Earth Planet Sci Lett 426:89– Chapman and Hall, New York514 pp. 100 Putnis A (1992) Introduction to mineral sciences . Cambridge Univ. Le Bas MJ, Le Maitre RW, Streckeisen A, Zanettin B (1986) A chemical Press, New York457 pp. classification of volcanic rocks based on the total alkali-silica dia- Rosi M, Sbrana A (1987) Phlegrean fields . CNR Quaderni de BLa – gram. J Petrol 27:745 750 Ricerca Scientifica^ 9, Rome175 pp. Lirer L, Luongo G, Scandone R (1987) On the volcanological evolution Rosi M, Sbrana A, Principe C (1983) The Phlegrean Fields: structural of Campi Flegrei. Eos Trans AGU 68 evolution, volcanic history and eruptive mechanism. J Volcanol Marsh BD (1988) Crystal size distribution (CSD) in rocks and the kinetics Geotherm Res 17:273–288 and dynamics of crystallization. Contrib Mineral Petrol 99:277–291 Schneider CA, Rasband WS, Eliceiri KW (2012) NIH image to ImageJ: Marsh BD (1998) On the interpretation of crystal size distributions in 25 years of image analysis. Nat Methods 9:671–675 magmatic systems. J Petrol 39:553–599 Shea T, Hammer JE (2013) Kinetics of cooling- and decompression- Martel C (2012) Eruption dynamics inferred from microlite crystalliza- induced crystallization in hydrous mafic-intermediate magmas. J tion experiments: application to Plinian and dome-forming eruptions Volcanol Geotherm Res 260:127–145 of Mt Pelée e (Martinique, Lesser Antilles. J Petrol 53:699–725 Swanson SE (1977) Relation of nucleation and crystal-growth rate to the Mastrolorenzo G, Pappalardo L (2006) Magma degassing and crystalli- development of granitic textures. Am Mineral 62:966–978 zation processes during eruptions of high-risk Neapolitan-volca- Vona A, Romano C, Dingwell DB, Giordano D (2011) The rheology of noes: evidence of common equilibrium rising processes in alkaline crystal-bearing basaltic magmas from Stromboli and Etna. Geochim magmas. Earth Planet Sci Lett 250:164–181 Cosmochim Acta 75:3214–3236